c Peter King, History of Philosophy Quarterly 1 (1984), 147165.

NSELMS Ontological Argument is an ad hominem argument against the Foole, part of which is a reductio ad absurdum, designed to prove the existence of God. The actual argument oered by St. Anselm has seventeen premisses; the heart of the argument is a careful distinction among intentional objectsand the Ontological Argument cannot be formalized by modal logic. It is not a modal argument at all, but rather relies on certain intuitive principles of intentional logic, which Anselm applies throughout the Proslogion. The Ontological Argument is valid, if one accepts these principles; insofar as an ad hominem argument may be sound, it is sound as well. It is not a demonstration, for the key premiss granted by the Foole is highly implausible. Those who agree with the Foole, however, may justiably assert Gods existence. These claims only apply to Anselms actual argument, not to other Ontological Arguments, no matter how distinguished the pedigree, no matter how careful the formalization. Other Ontological Arguments only interest me insofar as they shed light on, or claim to accurately represent, Anselms Ontological Argument. Other Ontological Arguments must be judged on their own merits. Anselms actual argument, unlike most versions, is an exercise in intentional logic, a fact that has eluded commentators from the time of Gaunilon. That Ontological Argument is the subject of this article, and henceforth I shall call it the Ontological Argument.
2. Anselms Actual Argument As is traditional, I call the argument presented in Proslogion 2 the Ontological Argument.1 This is not to deny that Anselm draws consequences from its premisses in the rest of the Proslogion, but that the argument of Proslogion 2 may be taken as a single argument, containing the conclusion
* I would like to thank Jim Van Aken for many valuable suggestions and conversations. This essay is dedicated to the memory of Jan Pinborg.
1

ANSELMS INTENTIONAL ARGUMENT*

1. Introduction

References to the Proslogion, Gaunilons Pro insipiente, and the Responsio are to the denitive critical text edited by F. S. Schmitt, Sancti Anselmi Cantuarensis archiepiscopi opera omnia Vol. 1 (Edinburgh: Nelson & Sons 19461961), 93139. All translations are my own. 1

ANSELMS INTENTIONAL ARGUMENT

announced in its title: That there really is a God. Our task now is to determine the principles Anselm uses.2 To do so, we shall carefully examine Anselms statement of his Ontological Argument. What follows is an exceedingly literal translation of Proslogion 2; the bracketed words have been added for sense.3 [S1] Therefore, O Lord, You Who give understanding to faith, give to me that I understand so much as You know to be t: that You are as we believe, and You are that which we believe. [S2] And, indeed, we believe You to be something than which nothing greater can be thought. [S3] Or [can it be that] there is not some such nature, then, since The Foole hath said in his heart: There is no God [Ps. 13:1]? [S4] But certainly, that same Foole, when he hears this very thing I say, something than which nothing greater can be thought, understands what he hears; and what he understands is in his understanding, even if he were not to understand that to be. [S5] It is one matter that a thing is in the understanding, another to understand a thing to be. [S6] For when the painter thinks beforehand what is going to be done, he has [it] in the understanding but does not yet understand to be what he does not yet make. [S7] Yet when he has painted, he both has [it] in the understanding and also understands to be what he now makes. [S8] Therefore,
2

The principles developed in the rest of this paper may be applied quite successfully to the rest of the Proslogion, in particular to the argument given in c. 3 and the claim common to c. 4 and c. 15 that God cannot be understood; for reasons of space I shall not consider such applications in this paper. Proslogion 2 (Schmitt 1.101102): [S1] Ergo, domine, qui das dei intellectum, da mihi, ut quantum scis expedire intelligam, quia es sicut credimus, et hoc es quod credimus. [S2] Et quidem credimus te esse aliquid quo nihil maius cogitari possit. [S3] An ergo non est aliqua talis natura, quia dixit insipiens in corde quo: non est deus? [S4] Sed certe ipse idem insipiens, cum audit hoc ipsum quod dico: aliquid quo maius nihil cogitari potest, intelligit quod audit; et quod intelligit in intellectu eius est, etiam si non intelligat illud esse. [S5] Aliud enim est rem esse in intellectu, aliud intelligere rem esse. [S6] Nam cum pictor praecogitat quae facturus est, habet quidem in intellectu sed nondum intelligit esse quod nondum fecit. [S7] Cum uero iam pinxit, et habet in intellectu et intelligit esse quod iam fecit. [S8] Conuincitur ergo etiam insipiens esse uel in intellectu aliquid quo nihil maius cogitari potest, quia hoc cum audit intelligit, et quidquid intelligitur in intellectu est. [S9] Et certe id quo maius cogitari nequit, non potest esse in solo intellectu. [S10] Si enim uel in solo intellectu est, potest cogitari esse et in re, quod maius est. [S11] Si ergo id quo maius cogitari non potest, est in solo intellectu: id ipsum quo maius cogitari non potest, est quo maius cogitari potest. [S12] Sed certe hoc esse non potest. [S13] Existit ergo procul dubio aliquid quo maius cogitari non ualet, et in intellectu et in re. c Peter King, History of Philosophy Quarterly 1 (1984), 147165.

3. THE PRELIMINARIES

even the Foole is convinced that there is in the understanding even something than which nothing greater can be thought, since when he hears this he understands, and whatever is understood is in the understanding. [S9] And certainly that than which a greater cannot be thought cannot be in the understanding alone. [S10] If indeed it is even in the understanding only, it can be thought to be in reality, which is greater. [S11] Thus if that than which a greater cannot be thought is in the understanding alone, the very thing than which a greater cannot be thought is [that] than which a greater can be thought. [S12] But certainly this cannot be. [S13] Therefore, without a doubt something than which a greater is not able to be thought exists (exsistit ), both in the understanding and in reality. The Ontological Argument is presented in these thirteen brief sentences. What is the argument? 3. The Preliminaries A standard a priori dismissal of the Ontological Argument is that it tries to do something impossible, namely argue from the conceptual to the real order, proving Gods existence from our concept of Him. But this is simply dogmatic. We can argue from the conceptual to the real order, and do so argue regularlyfrom our concept of round square we conclude that no round squares exist. Anselm proposes that in one very special case we can argue to the existence of something, namely Godand since the a priori dismissal of the argument is unwarranted, it is simple dogmatism to rule out his attempt a priori.4 A related objection is that Anselm denes God as existing; the illegitimacy of this is then pointed out: we may dene something to be an existent round square, but this hardly allows us to conclude that round squares exist. While this argument is correct, it is misplaced as a criticism of Anselm, for he does not dene God as existing. Indeed, if he were to so dene God, he would not need to give us an argument at all, merely propose and defend his denition. And that is not at all what takes place in Proslogion 2. In the statement of the Ontological Argument, [S1] is often taken to be a rhetorical ourish, a pious prater irrelevant to the main argument. It is not. In [S1] and [S2] Anselm is explicitly describing what the Ontological Argument proves: the existence of something than which nothing greater
4

The objection shows its weakness if we now revise it to claim that it is impossible to argue from the conceptual order to the existence of something, though we can so argue to prove nonexistence. What could possibly justify such a claim? c Peter King, History of Philosophy Quarterly 1 (1984), 147165.

ANSELMS INTENTIONAL ARGUMENT

can be thought, which by common consent is crucial to the notion of a deity. He does not say that this is a denition of God; all that the proof requires is that it be true of God. More precisely, he needs to hold that God is necessarily and uniquely something than which nothing greater can be thought. But that does not imply that God is dened as something than which nothing greater can be thought. God is also thought to be necessarily and uniquely omniscient, but nobody takes omniscience as a denition of God. Moreover, Anselm will justify the claim that God is something than which nothing greater can be thought in the rest of the Proslogion, by arguing that that than which a greater cannot be thought is also omniscient (c. 6), omnipotent (c. 7), merciful and just and good (c. 9 11), innite (c. 13), absolutely simple (c. 18), whose essence is existence (c. 12), and unique (c. 22)characteristics that do in fact properly belong to God alone. Thus the Ontological Argument of Proslogion 2 begins with a claim that is justied by his later investigation: (1) God is something than which nothing greater can be thought. We may accept (1) as plausible in itself, or accept it conditionally upon Anselms proving that something than which nothing greater can be thought is in fact omniscient, omnipotent, and so forth. Those who wish to dissent from (1) at this stage may ignore it and take the rest of the Ontological Argument to be solely a proof of the existence of something than which nothing greater can be thought. Nothing in the rest of the Ontological Argument turns on (1), and those who have objected to it by denying (1) are simply confused about the logical structure of the Ontological Argument. The Foole is introduced in [S3]. The Foole is central to the ad hominem nature of the Ontological Argument, for the Ontological Argument is based on the Fooles admission that there is something than which nothing greater can be thought in his understanding. But Anselm does not simply ask us to admit this along with the Foole; we get some reasons for admitting it. The rst step is explicitly stated in [S4] and [S8]: (2) Whatever is understood is in the understanding. Anselm oers a defense of (2) as follows:5 But you [the objector] will say that even if it is in the understanding, nevertheless [being in the understanding] does not follow because it is understood. Observe how it follows that it is in the understanding
5

Responsio 2 (Schmitt 1.132): Sed dices quia etsi est in intellectu, non tamen consequitur quia intelligitur. Vide quia consequitur esse in intellectu, ex eo quia intelligitur. Sicut enim quod cogitatur, cogitatione cogitatur, et quod cogitatione cogitatur, sicut cogitatur sic est in cogitatione: ita quod intelligitur intellectu intelligitur, et quod intellectu intelligitur, sicut intelligitur ita est in intellectu. Quid hoc planius? c Peter King, History of Philosophy Quarterly 1 (1984), 147165.

3. THE PRELIMINARIES

from this [fact], because it is understood. Indeed, just as what is thought is thought by means of a thought, and what is thought by means of a thought is thus in thought just as it is thought, so too what is understood is understood by the understanding, and what is understood by the understanding is thus in the understanding just as it is understood. What could be more clear than this? Anselm argues in this passage that (2) is a consequence of both the faculty of understanding and the exercise of that faculty (confusingly also called understanding). His point, though obscure, seems to be this: to understand something (as the exercise of a faculty) is to render that very thing itself understood, and to be understood is just to say that the thing understood is in the understanding. Thus (2) is a consequence of the nature of faculty of understanding. Whatever this obscure train of thought may be, we can extract from it the key point that Anselm is going to talk about the mind objectually, as though it were a realm inhabited by objects (in some sense of objects). This is as yet vague and imprecise. We shall clarify it in the next sections. Let us provisionally grant (2) and move to the next step of the argument. Although Anselm is not exact, his reasoning appears to be as follows. From [S3] we know that:6 (3) The Foole says that God does not exist. Now Anselm and the Foole have got to disagree, or there is no ad hominem argument; they must be talking about the same thing. Thus: (4) The Foole understands (1). Now to understand may take a direct object or a propositional complement; we may couch the discussion in terms of propositions for convenience. Nothing essential to the argument is lost of one chooses a dierent manner of expression. We need to add the plausible function-rule: (5) Understanding a proposition is a function of understanding its constituent parts. This need not be interpreted as a disguised semantic principle, or as concealing a belief that all propositions are categorical in form. A sentence cannot be understood unless the various parts of it are also understood. You cannot understand the sentence Jones is blictrix unless you understand what blictrix meansand this is true even allowing for understanding a word in context, for to understand a word in context simply is to have (some) understanding of the previously unfamiliar parts of the sentence. From (3)
6

Shifting from the quoted direct discourse of the Foole in [S3] to an indirect discourse for in (3) does not, I believe, alter the sense of the argument. c Peter King, History of Philosophy Quarterly 1 (1984), 147165.

ANSELMS INTENTIONAL ARGUMENT

(5) we deduce that Anselm and the Foole are in actual disagreement,7 and derive: (6) The Foole understands something than which nothing greater can be thought. From (2) and (6) we may then infer: (7) Something than which nothing greater can be thought is in the Fooles understanding. We could not proceed from (1)(4) to (7) without encountering problems with substitution in opaque contexts, but (5) and (6) allay that fear. 4. The Painter-Example Most commentators take [S4][S8] to have no more content than (1)(7). Yet if that were Anselms sole purpose, he could have accomplished it much more eciently. Instead, he talks about painters. Why? Look at [S4][S8] closely. The standard view takes the painter-example to be Anselms rambling medival way of distinguishing between something existing in the understanding and something existing in reality. The painter rst has the painting in his understanding; after he has painted, the painting is not merely in his understanding but also in reality. Such a distinction is certainly warranted after (2) is introduced, for we may then distinguish two dierent realms in which something exists, namely in the understanding and in reality. The burden of the Fooles objection is to make the distinction acute: what if God exists in the understanding alone, and not in reality? These observations are surely true, but incomplete. Anselm could easily have distinguished two realms in which something may exist immediately after (2) without any fuss about painters. Bur note that at the end of [S4] Anselm asserts that what the Foole understands is in his understanding, even if he were not to understand it to be. The phrase etiamsi non intelligat esse suggests that we take the Foole as conceiving God without conceiving God as existent. Indeed, this is what Anselm says in the very next sentence: It is one matter that a thing is in the understanding, another to understand a thing to be. He does not say that it is one matter that a thing is in the understanding, another that it is in realitywhich is all that the standard view requires. The painter-example holds the key to the Ontological Argument, for a careful reading will show that Anselm is drawing a distinction between objects of thought, the same but for one including existence and the other
7

Otherwise the Foole may retract his claim that God does not exist, which would be quite acceptable to Anselm. c Peter King, History of Philosophy Quarterly 1 (1984), 147165.

4. THE PAINTER-EXAMPLE

not. The relevant background for the Ontological Argument is not modal logic but various principles of intentional logic, for Anselm will be talking about the mind objectually and drawing a ne-grained distinction among such objects. The painter-example is introduced to make sure we do not miss this subtle distinction. (It didnt work.) An example will help motivate Anselms distinction. Consider Raphael as he is planning The School of Athens. Before he began to paint, he had some notion of what the nished product would look like; in the case of a genius like Raphael, we can even say that he knew exactly what the painting would look likethat he had The School of Athens in his understanding.8 Of course, he did not at that time understand to be what he does not yet make; imagining a painting is not enough; one must actually paint it. But what can we say after Raphael actually paints The School of Athens? On the one hand, he still has in his understanding the original conception of the nished product. Bur he now also has in his understanding The School of Athens as it is, as he has painted it on the walls of the Stanza della Segnatura. He knows his job is done, that he has painted the painting he conceived. At this point he both has [it] in the understanding and also understands to be what he now makes. Indeed, these things smust be distinct: otherwise it would be pointless to plan things in advance, for we could never compare the nished product to our original conception. Nor would it make any sense for Raphael to say e. g. Its not quite what I had in mind. This is the most natural reading of [S4][S8]. But there is further support found in the Responsio. Gaunilon had argued that in Anselms example the painting itself is not analogous to something than which nothing greater can be thought. Anselms reply to this objection is tart, pointing out what Gaunilon had overlooked (Responsio c. 8):9 But what you so carefully argue, that [that] than which something greater cannot be thought is not like the not-yet made painting in the painters understanding, misses the point. Indeed, I did not put forward the case of the painting known beforehand for this reason
8

Although the example is couched in terms of mental images, this is really irrelevant to the point Anselm is making. It is also true, but irrelevant to the example, that artistic creation probably does not proceed in the idealized fashion. Gaunilons objection is on Schmitt 1.126; Anselms reply at 1.137: Quod uero tam studiose probas quo maius cogitari nequit non tale esse qualis nondum facta pictura in intellectu pictoris: sine causa t. Non enim ad hoc protuli picturam praecogitatam, ut tale illud de quo agebatur uellem asserere sed tantum ut aliquid esse in intellectu, quod esse non intelligeretur, possem ostendere. c Peter King, History of Philosophy Quarterly 1 (1984), 147165.

ANSELMS INTENTIONAL ARGUMENT

I wanted to claim that it was like what was at issuebut rather so that I could show something not understood to be to be in the understanding. That is, Anselm was not drawing an exact analogy between the painting and something than which nothing greater can be thought, but making a distinction among types of items in the understanding. I shall henceforth call items in the understanding intentional objects, and the distinction Anselm will draw among them the intentional distinction. Much more needs to be said about both of these; Anselm says little directly, but much implicitly. In the next section I shall isolate the key philosophical features of his intentional-objects approach, and in 6 oer a defense of his use of intentional objects. 5. Intentional Objects: Analysis Anselm chooses to speak of objects in the understanding rather than talking of concepts. We shall as well. What is an intentional object? Fortunately, we do not have to give a thorough answer to this question. Here it suces to isolate some of their key features. First, as suggested above, intentional objects are mental items. They are not all of the contents of the mental realm; othersfeelings, desires, pains, and the likeare not intentional objects. Intentional objects are in some ways similar to what people have called incomplete objects, existing in the mental realm. An intentional object may be specied by any description,10 provided that the elements of the description are not mutually incompatible. Call this feature the internal selfconsistency of intentional objects. By elements not mutually incompatible I mean to exclude descriptions such as round and square or colored numberand even colorless green ideas. Later medival philosophers would call this the (mutual) repugnance of the termsthe repugnantia terminorum. The requirement of internal self-consistency, though, should not be taken to mean that intentional objects are precisely possible objects. On most modern understandings of modality, a possible object diers from an actual object only in inhabiting a nonactual possible world, or perhaps several nonactual possible worlds. Such a possible object, then, must be complete, that is, must possess the full complement of properties any object must have
10

Such a description may also specify existence or nonexistence. This will be defended later in this section and the next. For now, I use the neutral term characteristics to talk about the elements of such descriptions. c Peter King, History of Philosophy Quarterly 1 (1984), 147165.

5. INTENTIONAL OBJECTS: ANALYSIS

if it is to be actual. If a possible object is corporeal, for example, then it must also be colored. But intentional objects need not be complete in this way. An intentional object may be specied by a description not including all of the features required for an object to be actual. Such incomplete intentional objects cannot, of course, be actual unless further characteristics are added.11 If descriptions specify the characteristics of intentional objects, then it is also clear that the intentional object round-and-blue (i. e. specied by the description round and blue) may correspond to many actual items: a dyed basketball, a painted sphere. But the intentional object is not to be confused with any of these real items. Now such a description will include or exclude certain features: an intentional object is blue; it is not square; it is not light but heavy. Naturally, the intentional object must also possess features that are logical consequences of others it possesses, taken together. But the intentional object round-and-blue itself is not as it stands a possible object, for possible objects have all of the characteristics required to be actualand anything actually round and blue must also have a particular spatial location, a particular material composition, a particular weight and size, and so on.12 What of the intentional object round-and-blue-and-bronze? Anselm will want ne-grained distinctions among intentional objects, and to treat the intentional object round-and-blue-and-bronze as distinct from the intentional object round-and-blue. There are sucient grounds for distinguishing them, for one explicitly includes a characteristic, being bronze, the other lacks. Thus we may establish strict criteria for the identity of intentional objects: intentional objects are identical when they include the same characteristics,13 and otherwise distinct. Note that this distinction, the intentional
11

Most medival philosophers would at this point immediately distinguish intentional objects conceived with precision and without precision, that is, between intentional objects whose descriptions are taken to exclude any further specication of characteristics (conceived with precision) and those whose descriptions permit further characteristics to be added (conceived without precision). An intentional object whose description is not complete and is conceived with precision cannot, by denition, be actual. For the purposes of this article, we shall assume all intentional objects to be conceived without precision, that is, as admitting further specication. Nor will it do to identify an intentional object with sets of possible objects, namely identify e. g. the intentional object round-and-blue with the set of all possible objects that are round and blue, whatever other characteristics they may possess. This suggestion will be ruled out by the discussion in the succeeding paragraphs of the intentional distinction and of existence as a characteristic able to be included in such descriptions, and will be explicitly argued for in 6. This formula is meant to include characteristics that may not be mentioned in the specc Peter King, History of Philosophy Quarterly 1 (1984), 147165.

12

13

10

ANSELMS INTENTIONAL ARGUMENT

distinction, holds even if the intentional objects in question correspond to exactly the same actual or possible objects. One characteristic that may be included in the description of an intentional object is existence. Unless we permit this, we shall not be able to draw the distinction necessary for the Ontological Argument, a distinction between the intentional objects something-than-which-nothing-greatercan-be-thought and something-than-which-nothing-greater-can-be-thoughtas-existing. Thus we may specify distinct intentional objects by the descriptions round and square, round and square as existing, and round and square as not existing. No actual object can correspond to the third, and exactly the same objects will correspond to the rst and second. The plausibility of this is evident if we recall that intentional objects are mental items. For example, Snow Leopards are rapidly becoming extinct; I may wonder whether there are any left, and trudge o to Siberia in hopes of nding one. In this case I can think of a Snow Leopard without knowing whether there are anyfor ex hypothesi I do not know whether any exist, but I must be able to think of what they are like, for otherwise I wouldnt know if I have found one. And this is clearly dierent from thinking that there are Snow Leopards in the world, that they still exist and are not yet extinct. What goes for Snow Leopards in this example goes for our thinking generally: It is one matter that a thing is in the understanding, another to understand a thing to be. Not all of our thoughts are explicitly existential in character. More will be said about existence in the next section, but the example shows, I believe, that Anselms distinction is secure grounded in common experience. I have mentioned without much discussion that certain real objects may correspond to, or be identied with, certain intentional objects. I shall not try to give criteria for such identication, and only point out some features of the correspondence. First, a single intentional object may correspond to many real objects; the intentional object round-and-blue mentioned above was one such. Second, many intentional objects may correspond to a single real object; the intentional objects round and colored-and-round apply to the same objects, and some complete description d and d -as-existing correspond to exactly the same item(s). Finally, note that we may speak of an item as both existing in reality and in the understanding when there is an intentional object which corresponds to some real item. This is a handy abbreviation. Moreover, we might be tempted to identify the two, as Anselm does: there is nothing inherently absurd in saying that real objects
ifying description, but that are logical consequences of those that are so mentioned. c Peter King, History of Philosophy Quarterly 1 (1984), 147165.

6. INTENTIONAL OBJECTS: DEFENSE

11

are also intentional objects, just as there is nothing absurd in saying that real objects are possible objects, although in neither case does the converse necessarily hold. 6. Intentional Objects: Defense The strongest defense of my interpretation of intentional objects is that they seem to be presupposed by common sense. That is, each of the above claims is intuitively plausible. Nor should we worry about foisting a bizarre Meinongian ontology upon Anselm, for intentional objects are in a strong sense irrelevant to ontology: they exist in the understanding, but they do not exist in reality. And ontology is only concerned with reality, in its actual and possible forms. But clearly we can think of things. The theory of intentional objects as presented merely attempts to set out some of the common-sense principles we actually use when we talk about thinking of things. Indeed, it seems as though any adequate philosophical theory should be able to account for cases like the Snow Leopard example; if that is true, then some version of Anselms distinction should survive, and quite possible one strong enough for the Ontological Argument.14 In fact, although we have followed Anselms practice in speaking of intentional objects, it seems possible to drop out any reference to objects and proceed entirely in the linguistic mode: we could identify the descriptions that specify intentional objects as the very intentional objects themselves. The intentional distinction would then become a ne-grained linguistic distinction among descriptions, the existence of an intentional object (its internal self-consistency) would be simply the compatibility of the terms included in the description, and the correspondence-rules would be the criteria for the satisability of such descriptions. This linguistic reduction would accomplish two things: it would show that Anselm is not engaging an any bizarre ontology, countenancing any peculiar entities; it would also purge the Ontological Argument of its mentalistic avor. To be sure, more work needs to be done before such a theory would be acceptable in a rigorous
14

I would like to suggestno morethat the intentional-objects framework is also a plausible way to view Aristotelian philosophy of mind, in which the form in the mind is identical with the form in the (external) object; the identity in question is the correspondence of a real object to an intentional object. Anselm did not have Aristotles De anima, but he had the works of Boethius, which present an abridged form of Aristotles philosophy of mind. The theory of intentional objects is a useful way to construe many of Boethiuss scattered remarks, and it might even be very close to what Aristotle himself thought (although there is no consensus today about what Aristotle himself thought). c Peter King, History of Philosophy Quarterly 1 (1984), 147165.

12

ANSELMS INTENTIONAL ARGUMENT

philosophical fashion. But it is not implausible as it stands. Some philosophers will resist the claim that existence is a characteristic able to be included in the specication of an intentional object. Such resistance is probably based on worries about whether existence is a predicate, familiar to us from Kant and Moore. But such worries are misplaced.15 All Anselm needs for the Ontological Argument is some way of capturing the distinction between thinking of something as existing and thinking of it without bringing existence into the picture. And it seems undeniable that we in fact do this. I have purposely used the neutral term characteristic above so as to leave it open whether existence is a property on the same level as, say, blue, or should be understood as some kind of logical operator (e. g. as a quantier). The precise version of the intentional distinction Anselm needs for the Ontological Argument may be formulated as follows: (8) For any intentional object a that neither includes existence nor nonexistence, the intentional object a-as-existing is distinct from a. We have already seen grounds for accepting (8). But Anselm also needs a supplementary principle: (9) A person having an intentional object such as a described in (8) has or can have the distinct intentional object a-as-existing as well. Both (8) and (9) are implicit in Anselms discussion of the painter-example. And the intentional distinction given in (8) is the reason the Ontological Argument cannot be adequately represented in modal logic. For the standard method for formalizing the Ontological Argument is to take conceivability as translated by possibility.16 But such conceivability is wider than mere possible existence, for the standard translation takes a is conceivable as There is an a in some possible world W But this is precisely to blur Anselms careful distinction between conceiving
15

To forestall a possible misunderstanding, note that the claim that existence is not a predicate is taken to mean that, metaphysically, existence is not a property, and since the Ontological Argument argues that God possesses all perfections, which are a special sort of property, the Ontological Argument is awed. Such arguments are indeed awed, but they are not Anselms Ontological Argument. There is some justication for this; as we have seen in 5, Anselm does not allow impossible objects to be conceivable. But the requirement of internal self-consistency is not so strong as the requirement that the objects in question be possible objects, as opposed to being merely intentional objects; see the discussion in 56. c Peter King, History of Philosophy Quarterly 1 (1984), 147165.

16

8. THE REDUCTIO SUBPROOF

13

of something and conceiving of it as existing. Modal logic takes possible existence to be actual existence in some possible world. But we must allow for things conceived without the question of existence entering in. Perhaps some enriched form of modal logic is adequate to handle such a distinction, but the current ones are not. And this is why modal versions of the Ontological Argument are doomed to failure. 7. The Preliminaries Revisited Now that we have the intentional distinction of (8) rmly in mind, do we have to revise our understanding of (1)(7)? It may help to point out that Anselm has, in his ad hominem attack on the Foole in (3)(7), had the Foole grant that something than which nothing greater can be thought is an internally self-consistent intentional object. This suggests the following understanding of the phrase something than which nothing greater can be thought: (10) Let g be the intentional object specied by the description something than which nothing greater can be thought. For all intentional objects x, if x and g are distinct then g is greater than x. Thus we need not interpret something than which nothing greater can be thought as a simple denite description, which it manifestly is not; nor do we need to interpret it as a disguised modal quantier, quantifying across possible worlds. It is rather a quantication across all intentional objects. If we interpret something than which nothing greater can be thought as a quantication across all intentional objects as suggested in (10), then for the rst time we have some substantive content to the notion of greatness employed in the Ontological Argument. For the notion of greatness used in (10) is an ordering neither of actual objects nor of possible objects, nor yet of comparing objects across possible worlds, but rather of intentional objects. We shall have more to say about this in defense of (16). At the moment it suces to note that we take greatness to be a ranking, and that it is asymmetric, irreexive, and transitive, dening at least a partial ordering on the set of all intentional objects. To be sure, Anselm does not say this explicitly, but it is implicit in his reasoning. We shall return to (10) in 10, for, innocent as it appears, it contains the single aw of the Ontological Argument. 8. The Reductio Subproof In (1)(10) we have the form of a straightforward ad hominem argument. At this point Anselm adds a reductio subproof. First, we need to list all of
c Peter King, History of Philosophy Quarterly 1 (1984), 147165.

14

ANSELMS INTENTIONAL ARGUMENT

the ways a thing may exist or fail to exist, for the reductio is going to be in the service of disjunctive syllogism. Anselm does not add the list explicitly, but surely it is unobjectionable and what he had in mind: (11) Something may exist either (a ) in the understanding alone; (b ) in the understanding and in reality; (c ) in reality alone; (d ) in neither the understanding nor in reality. Since we are working with the ad hominem admission of (7), we may eliminate (11c) and (11d) for the case of something than which nothing greater can be thought. Thus we conclude: (12) Something than which nothing greater can be thought exists either (a ) in the understanding alone, or (b ) in the understanding and in reality. This, of course, is a restatement of the initial problem of [S3]. We want to show that (12b) is really the case, and we do so by disjunctive syllogism, by proving that (12a) cannot hold. The method of proof used to eliminate (12a) is the reductio ad absurdum ; both method and conclusion are stated in [S9]. It is worth pausing here for just a moment, for people are often confused about the structure of the Ontological Argument. The proof here is a reductio ; any old contradiction would suce for the argument, e. g. 12 + 5 = 13. The contradiction Anselm derives has to do with the nature of greatness. But he does not immediately conclude from this that something than which nothing greater can be thought exists, as most people seem to think; he concludes that (12a) is false and that (12b) must be the case. So the rest of the Ontological Argument is not a direct proof of the existence of God, or even of something than which nothing greater can be thought. This fact should be kept in mind. Even though the particular argument Anselm gives may not result in a contradiction, a close relative might, which is all he needs. Thus the next step of the proof, based on [S9], is: (13) Reductio -assumption: Assume (12a) is true, i. e. something than which nothing greater can be thought exists only in the understanding. With all of the machinery in place, we may now fearlessly confront the contradiction presented in [S10][S12]. 9. The Contradiction In order to derive the contradiction, which powers the whole argument, we need rst to distinguish two intentional objects. Let g be the intentional object specied by something than which nothing greater can be thought, which we know by (7) to be in the Fooles understanding. Now by (8) we
c Peter King, History of Philosophy Quarterly 1 (1984), 147165.

9. THE CONTRADICTION

15

can distinguish the intentional object from another intentional object, since g neither positively includes existence nor positively includes nonexistence: (14) The intentional object g is distinct from the intentional object g as-existing. But we need to add a restriction, since we are engaged in an ad hominem argument; we may derive it easily from (7) and (9), once we add the proviso that the Foole hears and understands Anselm as the Ontological Argument is set forth (which disposes of the possibility that although he could have the relevant intentional object in fact he does not, perhaps because he has never considered the matter): (15) The Foole has in his understanding the intentional object g -asexisting. This corresponds to the rst part of [S10], viz. If indeed it [something than which nothing greater can be thought] is even in the understanding only, it can be thought to be in reality. . . In the rest of [S10] Anselm says . . . which is greater. We interpret the quod here to refer to the intentional objects, a move indicated while discussing greatness in (10), and one certainly permitted by the rules of Latin grammar. Thus we have the following as the most natural reading of the text: (16) g -as-existing is greater than g . People often accuse Anselm of justifying (16) on the basis of another claim, namely: (16* )For any intentional object x, x-as-existing is greater than x. But Anselm never asserts (16*)! In fact, there is plenty of reason to think (16*) is false. There are many things whose existence is worse than their nonexistence: nuclear war, cancer, poverty, and so on. Nor does Anselm justify (16) by arguing as follows: (16a) Existence is a perfection. (16b) g -as-existing has a perfection g lacks, viz. existence. Therefore: g -as-existing is greater than g . This is better, but Anselm does not here claim that existence is a perfection. If he were to argue this way, we could equally conclude that cancer-asexisting is greater than cancer-as-not-existing. Yet even as a ranking of intentional objects I doubt this is trueand in the case of choosing between two possible worlds, in one which there is cancer and the other of which there is not (and alike in all other respects), we should have no hesitation in ranking the possible world without cancer as greater than the possible world with cancer. It is necessary to be very careful here, for we are using intuitions about
c Peter King, History of Philosophy Quarterly 1 (1984), 147165.

16

ANSELMS INTENTIONAL ARGUMENT

possible worlds to justify a particular ranking of intentional objects, and not all intentional objects are possible objects. Thus we cannot rank intentional objects simply by comparing possible worlds. How can we rank them? Anselm gives us no clue. The principle suggested in (16a) seems plausible, but it runs into counterexamples like the cancer case. Nor is it clear what (16a) would mean if directly applied to intentional objects. Why should we think that an intentional object possessing the characteristic of existing is any greater, qua intentional object, than the intentional object that does not include existence?17 Anselms failure to provide us with adequate principles for ranking intentional objects, that is, the lack of an analysis of greatness, is a serious omission in the Ontological Argument. Yet an omission may not amount to a logical aw: (16) may well be true, although we are not provided with a principled reason for thinking so. But Anselm makes no general claims here. He only asserts that the intentional object something-than-which-nothing-greater-can-be-thought-asexisting is in fact greater than the intentional object something-than-whichnothing-greater-can-be-thought. The claim about g and g -as-existing must stand or fall on its own merits.18 However, I am willing to believe that it is a good thing to have a being around than which nothing greater can be thought, even if the contrary were impossible. If it is a good thing, we may perhaps conclude that the intentional object g -as-existing is greater than the intentional object g . Let us therefore grant Anselms intuition and accept (16) as it stands, without committing ourselves to any of the dubious principles philosophers have suggested on Anselms behalf. We now come to the crux of the argument, [S11]: Thus if that than which a greater cannot be thought is in the understanding alone, the very thing than which a greater cannot be thought is [that] than which a greater can be thought. Anselms reasoning here is straightforward, now that we have done all the work. By (14) we know that g and g -as-existing are distinct intentional objects. By substitution of x = g -as-existing in (10) and modus ponens, we easily derive: (17) g is greater than g -as-existing. And since greatness is asymmetric, (16) and (17) contradict one another.
17

It could be argued that it is better even in the case of cancer, for then we do not think that certain deaths are caused by evil demons; thinking that cancer exists is at least a prerequisite for intelligently trying to nd a cure. But this seems a rather ad hoc justication of (16a). Indeed, we cant even pull the same trick as in the case of cancer by intuitively comparing possible worldsfor, as Anselm argues in c. 3, it is not possible that God not exist. c Peter King, History of Philosophy Quarterly 1 (1984), 147165.

18

10. WHY THE ARGUMENT REALLY DOESNT WORK

17

Therefore the reductio -assumption of (13) must be wrong, that something than which nothing greater can be thought exists only in the understanding, and so (12a) is false. By simple disjunctive syllogism we conclude (12b) something than which nothing greater can be thought exists in both the understanding and in reality. And by (1) we have our desired conclusion, using no more than simple rst-order logic in the reductio : Therefore: God exists. And that, as Anselm points out in [S12], is the conclusion of the Ontological Argument. 10. Why the Argument Really Doesnt Work The argument as a whole, then, looks like this: (1) God is something than which nothing greater can be thought. (2) Whatever is understood is in the understanding. (3) The Foole says that God does not exist. (4) The Foole understands (1). (5) Understanding a proposition is a function of understanding its constituent parts. (6) The Foole understands something than which nothing greater can be thought. (7) Something than which nothing greater can be thought is in the Fooles understanding. (8) For any intentional object a that neither includes existence nor nonexistence, the intentional object a-as-existing is distinct from a. (9) A person having an intentional object such as a described in (8) has or can have the distinct intentional object a-as-existing as well. (10) Let g be the intentional object specied by the description something than which nothing greater can be thought. For all intentional objects x, if x and g are distinct then g is greater than x. (11) Something may exist either (a ) in the understanding alone; (b ) in the understanding and in reality; (c ) in reality alone; (d ) in neither the understanding nor in reality. (12) Something than which nothing greater can be thought exists either (a ) in the understanding alone, or (b ) in the understanding and in reality. (13) Reductio -assumption: Assume (12a) is true, i. e. something than which nothing greater can be thought exists only in the understanding. (14) The intentional object g is distinct from the intentional object g c Peter King, History of Philosophy Quarterly 1 (1984), 147165.

18

ANSELMS INTENTIONAL ARGUMENT

as-existing. (15) The Foole has in his understanding the intentional object g -asexisting. (16) g -as-existing is greater than g . (17) g is greater than g -as-existing. Therefore: God exists. So whats wrong with the argument? Very little. Classical objections to (1) are beside the point, for it is not a denition. And (5) seems unobjectionable in the rather bloodless sense given to it in 3. The claims in (3) and (4) are concessions made by the Foole. The reductio -subproof of (11)(13) is a matter of simple logic. One might object to (2) and to (8)(10), but I have argued that they are not implausible. Certainly it might turn out that there are deep philosophical reasons to reject Anselms theory of intentional objects, but they must in fact be deep reasons: I have argued that the claims Anselm makes in this framework are prima facie plausible. There are two weak premisses. To take the easier case: no good support for (16) is oered by Anselm. Perhaps he thought that it was evident. It is not. A defense of (16) would require an analysis of greatness, which is conspicuously absent from the text. (You cant deny (16) because if you grant it the Ontological Argument works; that isnt fair.) No; (16) is a weak spot, but it is not implausible as it stands. And if Anselm is attempting to show the reasonableness of belief in God, that is all he needs. There is one genuine problem, and only one, with the Ontological Argument. It is (10) taken in conjunction with (6)(7). If the Foole allows (10) then the Ontological Argument worksas well as any argument ever works. But why does he not retract his claim in (4) when he hears and understands the construction placed upon something than which nothing greater can be thought in (10)? For the Foole might object to Anselm that one cannot quantify over the totality of intentional objects while forming an intentional object. After all, we know now that there are grave problems with quantication over a totality to dene a member of that totalitythe paradoxes of na ve set theory are witness to this. But this is a modern Fooles objection. A Foole of Anselms day might simply have pointed out that we have required intentional objects to be internally self-consistent, and it is not clear that something than which nothing greater can be thought species such a self-consistent entity. Anselm seems to be aware of this, and that is the deep reason why the Ontological Argument is an ad hominem argument: the Foole has, in (4), tacitly allowed
c Peter King, History of Philosophy Quarterly 1 (1984), 147165.

10. WHY THE ARGUMENT REALLY DOESNT WORK

19

that something than which nothing greater can be thought species such an internally consistent entity, i. e. an intentional object. And that is why Anselm is careful to insist in his Responsio that he was arguing against the Foole, and why in Responsio c. 1 he begins by appealing to Gaunilons faith and conscience as a Christian: Now my strongest appeal is to your faith and to your conscience. Anselm wants Gaunilon to make the same concessions as the Foole. Moreover, the logical structure of (10) is the reason why Anselm dismisses Gaunilons famous Lost Island Objectionthe objection that the Ontological Argument must fail because it proves too much, since it would equally show the existence of an island than which no greater island could be thought. Anselm merely remarks in Responsio c. 3 that the cases are not parallel. Why not? Because g is dened by quantifying across the totality of intentional objects, and the Lost Island by a limited and restricted quantication. The logic of the one claim is not the logic of the other, which is exactly what Anselm says:19 I wholeheartedly assert that if anyone were to nd for me anything either really existing or [existing] only in the understanding other than that than which a greater cannot be thought, to which the logic of my argument would apply, I shall nd the Lost Island and give it to him, never to be lost again. The Lost Island case is not parallel, for the denition of the Lost Island has a dierent logical structure, being a restricted quantication. But Anselm must get people to admit that something than which nothing greater can be thought is a coherent description. Anselm tries to make this claim plausible. That is why he argues that we can arrive at something than which nothing greater can be thought by a self-consistent procedure:20
19

Gaunilons Lost Island objection is given in Pro inspiente c. 6; the Latin text of Anselms reply in Responsio c. 3 is given on 1.133: Fidens loquor, quia si quis inuenerit mihi aut re ipsa aut sola cogitatione existens praeter quo maius cogitari non possit, cui aptare ualeat conexionem huius meae argumentationis: inueniam et dabo illi perditam insulam amplius non perdendam. Responsio c. 8 1.137: Item quod dicis quo maius cogitari nequit secundum rem uel ex genere tibi uel ex specie notam te cogitare auditum uel in intellectu habere non posse, quoniam nec ipsam rem nosti, nec eam ex alia simili potes conicere: palam est rem aliter sese habere. Quoniam namque omne minus bonum in tantum est simile maiori bono inquantum est bonum: patet cuilibet rationabili menti, quia de bonis minoribus ad maiora conscendendo ex iis quibus aliquid maius cogitari potest, multum possumus conicere illud quo nihil potest maius cogitari. . . Sic itaque facile refelli potest insipiens qui sacram auctoritatem non recipit, si negat quo maius cogitari non ualet ex aliis c Peter King, History of Philosophy Quarterly 1 (1984), 147165.

20

20

ANSELMS INTENTIONAL ARGUMENT

Again, you say that having heard that than which a greater cannot be thought [you] are not able to have it in your understanding as [an object] known to you in reality, either in particular or generally, since you neither know the thing itself not can arrive at it by similar things. But this is hardly the case, for surely anything less good is similar to a greater good, insofar as it is good; it is clear to any mind able to reason that we can conjecture much about that than which nothing greater can be thought by ascending from the lesser goods to the greater, from those things than which something greater can be thought. . . Thus in this fashion the Foole who does not accept Sacred Scripture can easily be refuted if he denies that he can arrive at [that] than which a greater cannot be thought from other things. But Anselm is wrong. I can know that one thing is hotter than another; I can form ideas about things hotter than those I have experienced by reecting on and extrapolating from my past experience of dealing with items hotter than others. But it makes no sense at all to speak of that than which nothing hotter can be thought. There is no superlative degree, no hottest thing; there is equally no reason to suppose on the basis of our experience with one thing greater than another that there is a greatest thing. What if greatness were measured ordinally? Then it would make no sense to speak of that than which nothing greater can be thought, for there is no number than which no greater number can be thought (natural numbers, that is). And there is no way to rule this possibility out of court. We are not forced to go along with Anselms Foole in granting (10). Anselms argument suers the same fate as all ad hominem arguments, namely it only works for those who make the crucial concessions. And we are not even tempted to do soor at least most of us are not. But those who go along with the Foole are correct to be convinced by the argument: it has no logical aws. 11. Conclusion The Ontological Argument is a better argument than commentators have thought; it has a complex structure relying on the intentional-objects framework. Its failure, if it can be said to fail at all, is die to the fact that most of us would not go along with the Foole in conceding that something than which nothing greater can be thought is a coherent description. Although there is no proof that it is not a coherent description, there are some reasons to think it is not. But these are merely reasons; they do not amount to a
rebus conici posse. c Peter King, History of Philosophy Quarterly 1 (1984), 147165.

11. CONCLUSION

21

disproof. We may criticize the Foole, but we cannot refute him. But it may be that criticism is sucient if Anselm is trying to establish the reasonableness of belief in Godand this must be his aim, for an ad hominem argument can do no more, not being a demonstration. We may undercut the whole Ontological Argument by showing the unreasonableness of the Fooles concessions. There are good reasons not to concede the claim that something than which nothing greater can be thought is as coherent description. Why, then, should one side with the Foole? Why, even, does the Foole concede it? Cur nisi quia stultus et insipiens? (Proslogion 4).

c Peter King, History of Philosophy Quarterly 1 (1984), 147165.